---
tags:
  - protein-sequences
datasets:
  - name: SCOPe-2.08
    type: csv
---

# Protein sequences from SCOPe-2.08

This dataset is based on section A.4.3 of [1]. For understanding the identification of remote homology relations in hidden layers, we consider the Astral SCOPe v2.08 dataset [2], containing genetic domain sequence subsets filtered to obtain < 40% pairwise sequence identity. Each domain is hierarchically classified into fold, super-family, and family. We impose an initial filter by excluding the Rossman-like folds (c.2–c.5, c.27 and 28, c.30 and 31) and the four- to eight-bladed b-propellers (b.66–b.70), as recommended in [3]. As a result, we obtained a dataset composed of 14535 sequences. For each specific task, we define an ad hoc dataset, which we will describe in detail in the corresponding sections A.5.2 and A.6.4, essentially to ensure sufficient population in the classes

## References
1. Lucrezia Valeriani, Francesca Cuturello, Alessio Ansuini, Alberto Cazzaniga. The geometry of hidden representations of protein language models. bioRxiv 2022.10.24.513504; doi: https://doi.org/10.1101/2022.10.24.513504
2. John-Marc Chandonia, Lindsey Guan, Shiangyi Lin, Changhua Yu, Naomi K Fox, and Steven E Brenner. SCOPe: improvements to the structural classification of proteins – extended database to facilitate variant interpretation and machine learning. Nucleic Acids Research, 50(D1):D553–D559, 12 2021
3. Johannes Söding and Michael Remmert. Protein sequence comparison and fold recognition: progress and good-practice benchmarking. Current opinion in structural biology, 21 3:404–411, 2011.

## Structure

- `full_data.csv` contains the details of 14535 proteins as decribed above .
- `dataset_script.py` contains the code to load and process the dataset.


## Usage

```python
from datasets import load_dataset

dataset = load_dataset("vkarthik095/filtered-SCOPe-2.08")